Summary¶
- I can get pretty results from this dataset after I processed the data, the best result is close to 74%
- Although the performance of these models are close, nerual network and decision tree would be faster than logistic regression a lot
- By features engineering, high blood pressure is the major factor which affects a person gets disease whether or not
- If to add more hidden layers, neural network could get a better performance
Introduction¶
Background¶
This dataset is to classify whether or not a patient has a certain unspecified disease according to his/her part of information. I used decision tree, logistic regression, multiple artificial neural networks and deep learning models to solve this problem. After validating model performance and tuning model.Finally, got the prediction for each model.
Dataset¶
Attributes’ information about the dataset (Disease Prediction Training.csv):
- Age: in years
- Gender: male/female
- Height: in unit of cm
- Weight: in unit of kg
- Low Blood Pressure: lower bound of blood pressure measurement
- High Blood Pressure: higher bound of blood pressure measurement
- Cholesterol: three cholesteral levels
- Glucose: three glucose levels
- Smoke: 1/0 regarding if the patient smokes
- Alcohol: 1/0 regarding if the patient drinks alcohol
- Exercise: 1/0 regarding if the patient exercises regularly
- Disease: The binary target variable. Does the patient have the disease?
Data Exploration¶
Package importing¶
In [1]:
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import plotly.express as px
from sklearn.preprocessing import StandardScaler
from sklearn import metrics
from sklearn.linear_model import LinearRegression, LogisticRegression
from sklearn.pipeline import make_pipeline
from sklearn.metrics import roc_curve, auc
from sklearn.model_selection import GridSearchCV, KFold,train_test_split,cross_val_score, cross_validate, ShuffleSplit, LeaveOneOut
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import confusion_matrix, classification_report
import time
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import GradientBoostingClassifier
np.random.seed(33)
Data Import¶
In [2]:
df = pd.read_csv("./Disease Prediction Training.csv")
In [3]:
df.head()
Out[3]:
Explore Data Analysis¶
Data Size and Types¶
In [4]:
df.shape
Out[4]:
In [5]:
df.dtypes
Out[5]:
Data Missing Value¶
In [6]:
df.isnull().sum()
Out[6]:
Data Value Analysis¶
In [7]:
df.describe()
Out[7]:
In [8]:
df.info()
In [9]:
df.head()
Out[9]:
Data Outlier Detection¶
In [10]:
ax = sns.distplot(df["Age"])
In [11]:
fig = px.violin(df, y="High Blood Pressure", x="Glucose", color="Gender", box=True, points="all",
hover_data=df.columns)
fig.show()
In [12]:
fig = px.violin(df, y="Low Blood Pressure", x="Glucose", color="Gender", box=True, points="all",
hover_data=df.columns)
fig.show()